Transitions to mobility difficulty associated with lower extremity osteoarthritis in high functioning older women: Longitudinal data from the Women's Health and Aging Study II

Authors


Abstract

Objective

To assess the impact of lower extremity osteoarthritis (OA) on transitions to mobility difficulty, and to assess the influence of pain, excess weight, and quadriceps strength on these transitions.

Methods

We analyzed longitudinal data acquired from 199 participants in the Women's Health and Aging Study II (ages 70–79 years) who initially reported no lower extremity limitation (e.g., difficulty walking one-quarter mile) or difficulty in activities of daily living (ADL; e.g., transferring). Prevalent lower extremity OA was determined from validated algorithms encompassing multiple data sources. Markov transition models were created to analyze the first transition from no difficulty at baseline to lower extremity limitations, ADL difficulty, or both 18, 36, and 72 months later.

Results

Compared with women without OA (n = 140), a higher proportion of women with lower extremity OA (n = 59) initially reported pain on most days and more severe pain while walking (P < 0.05). Women with OA were also heavier, with a higher proportion being obese or overweight (P < 0.001). Lower extremity OA, higher body mass index, and lower knee extensor strength independently increased the risk of transition to combined lower extremity and ADL difficulty first over 72 months.

Conclusion

Lower extremity OA increased the likelihood of developing difficulty in both lower extremity tasks and ADL over 72 months in a cohort of initially high functioning older women. Two modifiable factors, higher relative weight and lower knee extensor strength, substantially impacted these transitions, and therefore warrant increased attention in the management of lower extremity OA.

INTRODUCTION

Osteoarthritis (OA) of weight-bearing joints is highly prevalent among older adults (1) and is strongly linked to late-life disability (2–4). However, the process through which OA-associated disablement develops has not been well characterized. Cross-sectional investigations, including one from the Women's Health and Aging Study II (WHAS II) (5), clearly demonstrate that knee OA is associated with painful symptoms, excess weight or obesity, muscle weakness, mobility difficulty, and slow mobility-related task performance (5–11). However, these observations have originated from samples of patients or research participants in which functional limitations often coexist with pain, obesity, and weakness, and therefore we cannot discern whether these factors contribute to or result from limited function. Using longitudinal data obtained from participants in the WHAS II who initially reported no mobility difficulty, we tested the hypothesis that women with lower extremity OA are more prone to develop mobility limitations than women without OA, and that limitations in lower extremity mobility tasks develop before difficulty in activities of daily living (ADL). We tested the additional hypothesis that pain, muscle strength, and body mass index (BMI) mediated these transitions over 72 months.

SUBJECTS AND METHODS

Design.

WHAS II is a prospective, population-based observational study designed to elucidate the transition from well functioning to preclinical and clinical disability in community-resident older women.

Study population.

This study was approved by the Institutional Review Board of the Johns Hopkins Medical Institutions. The methodology for screening and recruitment of the WHAS II sample has been published previously (12–14). In brief, the WHAS II sample was drawn from the Health Care Financing Administration Medicare files that list all female beneficiaries ages 70–79 years in 12 contiguous zip code areas in Baltimore City and County, Maryland. Women were eligible to participate on the basis of a screening interview if they scored ≥24 on the Mini-Mental State Examination (15) and reported difficulty with tasks in no more than 1 of 4 functional domains encompassing mobility, upper extremity function, and instrumental and basic self-care activities. Of 1,630 women screened, 880 met the eligibility criteria and 436 (49.5%) agreed to participate in the extensive in-clinic examination. Women who were eligible but declined participation were less educated, had lower income, and were more likely to rate their health status as fair or poor (16). As shown in Figure 1, the final study sample comprised women who reported no difficulty at baseline and had complete covariate information available for analysis. Women classified as having definite rheumatoid arthritis or possible OA of the knee or hip were excluded from the analysis.

Figure 1.

Study sample. WHAS II = Women's Health and Aging Study II; OA = osteoarthritis.

Data collection.

Study participants were evaluated in the Johns Hopkins Functional Status Laboratory by trained evaluators. These evaluators used standardized protocols that encompassed interviewer-administered questionnaires, a comprehensive physical examination, and strength testing at baseline and 18, 36, and 72 months after the initial evaluation (12–14).

Osteoarthritis status.

Participants were categorized as having “lower extremity OA” or “no OA” using published algorithms (16) modeled according to the American College of Rheumatology classification criteria (17). Women were classified as having OA of the knees and/or hips on the basis of a reported physician diagnosis of arthritis specified as either OA or degenerative arthritis. The diagnosis was confirmed by physician questionnaire, medical record review, and/or clinic examination criteria including knee tenderness on palpation, painful or limited passive motion, deformity, crepitus and bony enlargement, and painful or limited motion of the hips, and radiographic features of OA when available (18). Women who reported no physician diagnosis of either knee or hip arthritis and whose lack of arthritis was confirmed negative on physician questionnaire or medical record review and physical examination were categorized as having “no OA.”

Assessment of hypothesized intermediary variables.

Pain presence was determined from the following question: “During the past year, have you had pain, aching, or discomfort in your knees/hips on most days for at least 1 month?” Items from the Western Ontario and McMaster Universities Osteoarthritis Index (19) were used to assess pain severity. Participants were asked to rate the average pain they experienced when performing specific activities (e.g., walking) using a 1–10-point scale. Body weight (kg) was measured with a calibrated medical scale. Height was measured using a standiometer to the nearest centimeter. BMI was calculated and categorized as <25 kg/m2, 25–29 kg/m2, and ≥30 kg/m2. Maximal isometric strength of the knee extensor muscles was measured, in kilograms, using a hand-held dynamometer (Nicholas Manual Muscle Tester, model #BK-7454; Fred Sammons, Burr Ridge, IL) (20, 21). During testing, participants were strongly encouraged to gradually increase the force to the greatest possible level while the tester was opposing. The maximum strength of 2 trials was used for each knee unless only 1 was available, and the average of left and right knee strength was used in the analyses. The interrater reliability for strength testing achieved an intraclass correlation coefficient of 0.91 as previously published (22). As an alternative measure, knee torque was calculated as [knee strength (kg) × 9.81 × knee height (meters) × cos(150)]/body weight (kg) (20, 21).

Assessment of mobility difficulty.

Participants were assessed at baseline and 18, 36, and 72 months following the initial evaluation. To assess task difficulty, participants were asked, “By yourself, that is without help from another person or special equipment, do you have any difficulty (walking a half mile, walking up 10 steps, lifting or carrying something as heavy as 10 pounds, getting in and out of a car or bus, getting in and out of bed or chairs, dressing, or bathing or showering)?” and “Because of a health or physical problem, do you have any difficulty (doing heavy housework or shopping for personal items)?” We defined incident difficulty as newly reported difficulty with any of the aforementioned tasks that was not previously reported at baseline. To facilitate analysis of disability transitions, we used previously published constructs of functional limitations (22): lower extremity mobility (walking a half mile, walking up 10 steps, doing heavy housework, or lifting or carrying something as heavy as 10 pounds) or ADL (getting in or out of bed or chairs, dressing, toileting, bathing). We defined combined lower extremity limitations and ADL difficulty as difficulty reported for ≥1 tasks within each category.

Covariates.

Questionnaires administered during the interview were used to determine participant age (years), race, marital status, living arrangement, depressive symptoms using the Geriatric Depression Scale (23), and medications used (24). Included in the final analysis was the number of adjudicated “definite” conditions including coronary artery disease (i.e., angina pectoris and/or myocardial infarction), congestive heart failure, peripheral arterial disease, degenerative disc disease, spinal stenosis, hip fracture, stroke, Parkinson's disease, pulmonary disease, diabetes mellitus, osteoporosis, and cancer, all of which were ascertained using standardized algorithms (25) that incorporated self report of physician diagnoses, medication use, physical examination, physician questionnaires, and/or a thorough medical record review.

Statistical analysis.

Analysis of variance for continuous variables and chi-square test for categorical variables were used to compare the baseline characteristics and mobility outcomes of women with lower extremity OA with that of women with no OA in the 199 women with complete covariate information who initially reported no mobility difficulty. To test the hypothesis that lower extremity OA results in lower extremity limitations first followed by ADL difficulty, first-order Markov models (26, 27) were used to analyze the transitions from no mobility difficulty at visit k-1 to 1 of the 3 mutually exclusive difficulty categories (i.e., lower extremity, ADL, and combined) at visit k, with k = 2, 3, and 4 representing the 3 followup visits at 18, 36, and 72 months after the baseline evaluation. Because our focus was on the first incident event of the 3 types, all subsequent transitions after the first event were excluded from the analyses. The parameters of the Markov models were estimated using polytomous logistic regression (28). Three longitudinal analysis models were implemented with adjustment for age, years of education, and the comorbid conditions specified above. Model I examined the main effect of OA. Model II added baseline pain, BMI category, and knee extensor strength to assess whether these were independently associated with transition to greater difficulty, and their roles as potential mediators of these transitions. In model III, knee torque replaced BMI and knee extensor strength. Analyses were performed using SAS version 8.2 (SAS Institute, Cary, NC) and STATA version 9 (STATA, College Station, TX).

RESULTS

To test our hypothesis, we focused the analysis on the subset of 199 women who reported no difficulty with mobility tasks at the time of the initial evaluation and had complete covariate information. This included 59 women classified as having OA of the lower extremity (44 women with knee OA only, 5 with hip OA only, and 10 with both knee and hip OA) on the basis of the WHAS II adjudication algorithms. The “no OA” group was comprised of 140 women classified as having neither knee nor hip OA.

As shown in Table 1, the OA group was comparable with the no OA group with regard to age, education, race, and number of adjudicated comorbid diseases. A greater proportion of women in the OA group reported use of arthritis medications than women in the no OA group (20.3% versus 9.4%; P < 0.05). Despite this, a higher proportion of women with OA reported pain on most days (57.6% versus 24.3%; P < 0.05) and greater pain severity while walking and climbing stairs (P < 0.05) than women in the no OA group. Women with OA were also heavier than women with no OA (mean ± SD BMI 27.3 ± 4.5 kg/m2 versus 25.1 ± 4.4 kg/m2; P = 0.01). Furthermore, a higher proportion of the OA versus the no OA group were obese (26.4% versus 10.89%; P = 0.02). An additional 35.9% of women in the OA group were overweight. The groups did not differ significantly with regard to knee extensor muscle strength (P = 0.44) or torque (P = 0.24).

Table 1. Baseline characteristics of Women's Health and Aging Study II participants by lower extremity osteoarthritis (OA) category (n = 199)*
VariablesLower extremity OA (n = 59)No lower extremity OA (n = 140)P
  • *

    Values are the mean ± SD unless otherwise indicated.

  • The number of adjudicated “definite” conditions other than OA.

  • Report of “pain, aching, or discomfort in your knees on most days for at least 1 month.”

  • §

    Western Ontario and McMaster Universities Osteoarthritis Index score.

  • Knee torque = [knee strength (kg) × 9.81 × knee height (meters) × cos(150)]/weight (kg).

Age, years73.9 ± 3.073.4 ± 2.70.23
School grades completed, years12.6 ± 3.113.2 ± 3.10.22
White race, %79.786.40.23
Number of diseases0.7 ± 0.70.8 ± 0.80.54
Pain presence,%57.624.3< 0.001
Average pain on walking§2.3 ± 2.51.8 ± 2.4< 0.01
Arthritis medication use, %20.39.40.03
Knee strength, kg20.3 ± 5.319.6 ± 5.70.44
Knee torque, Newtons2.93 ± 0.653.09 ± 0.940.24
Body mass index, kg/m227.3 ± 4.525.1 ± 4.4< 0.01
  <25, %37.755.50.02
  25–29, %35.933.6 
  ≥30, %26.410.9 

As shown in Table 2, ∼40% of women reported no task difficulty transitions, and 30% reported transitions to lower extremity limitations first across both groups. Although we observed no difference between the OA and no OA groups in incident mobility transitions based on a global test, comparison of specific transitions revealed that a higher proportion of women with OA transitioned to combined lower extremity limitations and ADL task difficulty first than did women without OA (22% and 10%, respectively). Conversely, transitioning to ADL difficulty first was reported by fewer women with OA than by women with no OA (8.5% and 14.3%, respectively). We then categorized women with and without lower extremity OA by presence or absence of pain, aching, or discomfort on most days for at least 1 month (Table 3). This categorization enhanced the strength of the association between lower extremity OA and the initial transition to combined lower extremity limitation with ADL difficulty in women who reported pain compared with women who did not report pain. Women with lower extremity OA and pain were therefore also less likely to transition to lower extremity limitations alone compared with women without OA but who reported pain. In contrast, women with OA but no pain were more likely to transition to lower extremity limitation alone compared with women with neither OA nor pain.

Table 2. Frequencies of initial transitions to task difficulty by initial OA classification*
TransitionLower extremity OA (n = 59)No lower extremity OA (n = 140)P
  • *

    Values are the number (percentage) unless otherwise indicated. OA = osteoarthritis; ADL = activities of daily living.

No new difficulty25 (42.4)64 (45.7)0.1235
Lower extremity only16 (27.1)42 (30.0) 
ADL only5 (8.5)20 (14.3) 
Lower extremity and ADL13 (22.0)14 (10.0) 
Table 3. Frequencies of initial transitions to task difficulty by initial OA classification and presence of pain*
TransitionLower extremity OA (n = 59)No lower extremity OA (n = 140)P
PainNo painPainNo pain
  • *

    Report of “pain, aching, or discomfort in your knees on most days for at least 1 month.” Values are the number (percentage) unless otherwise indicated. OA = osteoarthritis; ADL = activities of daily living.

No new difficulty16 (47.1)9 (36.0)16 (44.4)48 (46.2)0.0812
Lower extremity only6 (17.6)10 (40.0)9 (25.0)33 (31.7) 
ADL only3 (8.8)2 (8.0)9 (25.0)11 (10.6) 
Lower extremity and ADL9 (26.5)4 (16.0)2 (5.6)12 (11.5) 

The results of the multivariate analyses are summarized in Table 4. As shown in model I, women with lower extremity OA at baseline were ∼2.5 times more likely than women with no OA to develop difficulty in both lower extremity and ADL difficulty over 18–36 months (odds ratio [OR] 2.48, 95% confidence interval [95% CI] 1.08–5.67; P < 0.05). The addition of pain to the OA model did not significantly alter the effect of OA on these transitions (data not shown). After adjusting for BMI and strength, pain appeared to reduce the risk of transition to lower extremity limitations first (OR 0.49, 95% CI 0.25–0.99), but was not significantly associated with transitions to the other mobility categories. The results were consistent whether pain presence or severity was used in the models. Furthermore, the likelihood ratio test was not significant for an interaction between pain presence and OA for predicting disability transitions (data not shown). This association between OA and transition to combined lower extremity limitations with ADL difficulty appears to be partly explained by BMI, given the increased risk observed in model II (OR 1.18, 95% CI 1.06–1.31). Only higher knee extensor strength reduced the risk of transition to ADL difficulty (OR 0.91, 95% CI 0.84–0.99; model II), independent of OA. Finally, we observed that greater knee torque reduced the risk of transition to difficulty in ADL with concurrent transition to lower extremity limitation (OR 0.53, 95% CI 0.30–0.94) and without concurrent transition to lower extremity limitation (OR 0.58, 95% CI 0.35–0.97), but did not reduce the risk for lower extremity difficulty alone (OR 0.74, 95% CI 0.52–1.06).

Table 4. Estimated odds ratios (95% confidence interval) of longitudinal transition to difficulty categories in women with no difficulty at baseline (n = 199)*
 Model IModel IIModel III
Both vs. noneLEL vs. noneADL vs. noneBoth vs. noneLEL vs. noneADL vs. noneBoth vs. noneLEL vs. noneADL vs. none
  • *

    Adjusted for age, education, race, and number of adjudicated conditions. vs = versus; LEL = lower extremity limitation; ADL = activities of daily living; OA = osteoarthritis; BMI = body mass index.

  • P <0.05.

  • Pain is “pain, aching, or discomfort in your knees on most days for at least 1 month.”

  • §

    Calculated as body weight (kg)/height (m)2.

  • P <0.01.

  • #

    Isometric knee extensor strength (kg).

  • **

    Torque was calculated as [knee strength (kg) × 9.81 × knee height (meters) × cos(150)]/body weight (kg).

OA2.48 (1.08–5.67)0.95 (0.50–1.81)0.60 (0.22–1.68)1.84 (0.73–4.64)1.13 (0.55–2.29)0.49 (0.16–1.48)2.43 (1.01–5.87)1.16 (0.58–2.32)0.48 (0.16–1.43)
Pain   0.97 (0.38–2.47)0.49 (0.25–0.99)1.64 (0.38–3.96)0.83 (0.33–2.07)0.49 (0.24–0.98)1.58 (0.66–3.81)
BMI§   1.18 (1.06–1.31)1.05 (0.98–1.13)1.04 (0.93–1.15)   
Strength#   0.94 (0.86–1.02)0.95 (0.90–1.01)0.91 (0.84–0.99)   
Torque**      0.53 (0.30–0.94)0.7 (0.52–1.06)0.58 (0.35–0.97)

DISCUSSION

We tested the hypothesis that limitations in lower extremity mobility tasks develop before difficulty in ADL, and we assessed the contributions of pain, muscle strength, and relative body weight to these transitions over 72 months using data obtained from WHAS II participants selected to represent the two-thirds highest functioning, community-resident women ages 70–79 years. Contrary to this hypothesis, the results suggest that lower extremity OA increases risk of transition to difficulty with both lower extremity mobility and ADL tasks simultaneously. Importantly, the results also suggest that the transition to greater difficulty in both task categories appears to be mediated by higher body mass, and they illustrate the protective effects of higher knee extensor torque in reducing risk of this transition as well as transition to ADL difficulty alone, independent of OA status.

Few studies have focused on the long-term functional impact of OA. The First National Health and Nutrition Examination Survey (NHANES-I) epidemiologic followup study provided the first evidence that symptomatic knee OA increases risk of self-reported difficulty with specific mobility tasks 10 years following the initial evaluation, and that the risk of difficulty is amplified by obesity (3, 29). More recently, investigators from the Observational Arthritis Study in Seniors trial (30) and the Mechanical Factors in Arthritis of the Knee study (31) have independently demonstrated that greater knee pain intensity and lower knee extensor strength increase risk of poor function in disabled older adults with knee OA. Despite numerous cross-sectional reports of the associations between OA, painful symptoms, low strength, and obesity, existing studies have not been able to discern the relative contributions of these closely related factors to the development of specific mobility difficulty characteristics. Since the development of mobility difficulty in older adults has been successfully mapped using transition models (32, 33), this method of analysis would also lend itself to examining the longitudinal effects of OA and OA-associated characteristics of pain, strength, and obesity on mobility function over time.

Our bivariate results demonstrate that lower extremity OA is associated with painful symptoms, excess weight, and obesity. Painful symptoms are consistently associated with poor function across numerous studies, including WHAS II (5, 7, 10, 11), and are also prevalent in this subset of women with OA who reported no difficulty at baseline, despite use of arthritis medications. However, these longitudinal results suggest that women with OA and pain are more likely to transition to combined lower extremity limitations with ADL difficulty and are less likely to transition to lower extremity limitations alone compared with women with asymptomatic OA. These results should not be misinterpreted to suggest that pain is not an important clinical target, but the results do help us understand the contribution of pain to mobility transitions in the context of excess body weight and knee extensor strength.

The observation that the transition to greater difficulty in both task categories associated with lower extremity OA appears to be partially mediated by higher body mass is a novel contribution to the existing literature. Although numerous cross-sectional reports have linked excess weight and obesity to knee pain (34) and to bilateral knee OA development and progression (35–38), few studies have examined the contribution of obesity and excess weight to functional impairments in the context of OA. In the NHANES-III, women with a BMI ≥30 or who were in the highest quintile of body fatness were twice as likely to report functional limitations as women of ideal body weight (39). Similarly, Visser and colleagues provide longitudinal evidence that high fat mass, assessed by bioelectric impedance, increased the risk of disability development 3 years following the initial evaluation in the Cardiovascular Health Study (40). Although arthritis was addressed as a covariate in both of these studies, arthritis assessment was limited to self report.

It is well accepted that lower knee extensor strength and poor mobility function are associated with OA of the knee (5–7, 41, 42). In contrast to these published observations, which include our WHAS II cross-sectional observations, we did not observe an association between OA and lower knee extensor strength in the subset of women without mobility difficulty at baseline. The multivariate findings that higher knee torque, but not extensor strength, confers protection from transition to lower extremity limitations and ADL task difficulty, as well as ADL difficulty alone imply the importance of considering knee extensor strength relative to body size and weight. Because these measures have been successfully applied during a home-based examination as previously reported (21, 22), the methods used to assess strength and torque could be applied to clinical setting.

Longitudinal observations of well-functioning women living in the community who are at risk of functional decline provide a unique perspective from which OA can be examined. The use of validated state-of-the-art algorithms to classify subjects with knee and hip OA and other chronic conditions is an additional strength and allows adjustment for comorbidities and other potential confounding factors. The authors acknowledge the following limitations of this study. First, radiographic data were not available for all subjects. However, the American College of Rheumatology classification system for OA (17) provides the basis for the validated algorithm by which participants were classified and incorporates multiple sources of clinical information. Second, the lengthy interval between observations makes it difficult to determine the true extent to which lower extremity limitation may precede ADL difficulty, and may explain why we did not observe the disablement sequence originally hypothesized in association with OA. It is also possible that the results may have been influenced by sampling bias towards underestimating transition risks due to the lower proportion of women participating who reported their health as poor, and due to the inclusion of a sample enriched with women less likely to report poor health and disability. Finally, although our analysis included depression score, we acknowledge that transitions to greater difficulty may be influenced by psychological factors that were not assessed in this study (i.e., mobility-specific self efficacy), the importance of which has been demonstrated in other longitudinal studies of physical function in disabled adults with knee OA (31, 43).

Importantly, this study emphasizes the functional consequences of lower extremity OA and of excess weight and the protective effect of higher knee extensor torque over a 72-month observation period in a cohort of initially well-functioning older women. Fortunately, recent intervention trials provide evidence that self-reported mobility impairment is amenable to intervention with modest weight loss and exercise in overweight or obese adults with knee OA (44). The findings of this study are relevant to growing numbers of Americans who are disproportionately female, aging, and increasingly obese. From 1960 to 1999, the prevalence of overweight adults increased from 44% to 61%, and the prevalence of obesity doubled from 13% to 27% (45–47) and has reached epidemic proportions in the US (48). The trends towards earlier onset of obesity observed between 1991 and 1998 (49) would predictably translate into a higher proportion of adults who could develop painful symptoms and mobility difficulty at an earlier age (50).

Acknowledgements

We wish to acknowledge the Women's Health and Aging Study group for their contributions.

Ancillary